Survey of Research on Artificial Intelligence and Cybersecurity

In recent years, the rapid development of artificial intelligence technology has brought significant impetus to social improvement. Meanwhile, there are also many interwoven problems between artificial intelligence(AI) and cybersecurity. Studying the security issues arising from the integration of artificial intelligence and cybersecurity is of great significance. This paper starts from the history of the development of artificial intelligence technology, combines the current international and domestic cybersecurity status and trends, and summarizes the current research progress in the field of AI security from three aspects: AI enabling security, AI endogenous security, and AI derived security. In the part of AI enabling security, Starting from the implementation of machine learning technologies in cybersecurity, this paper sums up the application of deep reinforcement learning methods in the field of cybersecurity from traditional machine learning and deep learning to those emerging in recent years; For the analysis of AI endogenous security, the research focus and key issues of AI endogenous security are elaborated from four aspects: AI base security, AI data security, AI algorithm security and AI model security, and provide our opinions and ideas based on our research accordingly; In the field of AI derived security, a brief summary of its connotation and research status is made. Finally, this paper provides an in-depth analysis of the trends in development and challenges of artificial intelligence and cybersecurity based on relevant research in the past decade, further analyzes the possibility of interdisciplinary research, and tries to put forward a new understanding of the connotation and extension of AI security, aiming to provide guidance for the future development direction of this field.